The K-Means algorithm has the ability to include pre-defined "fixed" cluster positions. You might use this facility if you were using MPCluster to find locations for new distribution depots, in addition to a set of existing depots.
Enable this option by setting the Include Fixed Clusters checkbox on the main dialog box:
Then press the Set Fixed Clusters button to display the Pre-existing (Fixed) Cluster Locations dialog box:
Use this dialog box to set the fixed cluster positions and parameters. First, set the Input Data. This is a Maptitude point data view or selection set that contains the fixed cluster positions. MPCluster will scan the selection and report the number of fixed clusters (i.e. data points) that were found.
You should then select the Name/Identifier Field from this input data. This field contains the name values that will be used in MPCluster's output, to identify these fixed clusters. A fixed cluster's name will be "Fixed:" followed by the name value in the Name/Identifier Field. The non-fixed clusters will receive automatic names such as "cluster1", "cluster2", etc.
Three constraints can be set. These are all maximum constraints that limit the eventual size of the fixed clusters. These are: Maximum number of points per cluster; Maximum cluster diameter (or radius); and Constraint sum less than. These all work in the same manner as the regular cluster options. The Constrain sum of setting uses the Constrain sum of data field that is enabled on the main panel, and is only enabled if the main panel setting has been enabled.
There are no minimum constraints because the fixed cluster positions will always be reported as valid cluster positions - regardless of whether there is sufficient data nearby. Therefore, a minimum constraint would be effectively ignored.
Press the OK button to save these settings and return to the main panel.